Our previous studies (presented in Preliminary Results) have shown that multivariate analysis of transcripts related to inflammatory response and to cellular stress can distinguish Alzheimer's disease peripheral blood leukocytes from control leukocytes. This ability to distinguish clinically diagnosed AD from controls on the basis of expression profile of easily obtained samples raises the question; can we distinguish preclinical AD on the basis of gene expression profiles of peripheral blood leukocytes? To answer this question requires obtaining blood samples from a cohort of non-demented persons then following them longitudinally for years to detect new AD cases in the cohort. The cohort may be enriched by selecting persons at risk, as was done in the ADAPT study which defined risk on the basis of age and of AD in a first degree relative. In the study proposed here we enrich our cohort of non-demented persons at risk for AD on the basis of carrying one of the well-defined familial gene mutations that lead, inevitably, to subsequent AD. In this proposal we focus largely on the mutated gene for PS1 since it is a frequently mutated gene among the FAD gene mutations. The central hypothesis to be tested by the research proposed here is, then; that those persons with an FAD mutation but diagnosis of no AD represent "preclinical" AD and that they will present a gene expression profile that will be predictive of subsequent AD by virtue of being similar to the AD gene expression profile of family members with the same mutation and with AD. We will extract RNA from peripheral blood leukocytes obtained from three classes of persons: 1) with an FAD mutation and with diagnosed AD, 2) with an FAD mutation and diagnosis of no AD and 3) wild type with no FAD mutation and a diagnosis of no AD. Where possible, these three classes will be derived from the same family. Gene expression in the extracted RNA will be examined by quantitative RT-PCR (qRT-PCR) with the targets being those transcripts related to inflammatory response and cellular stress that we have defined in our earlier work. Statistical analysis of these data will be by means of multivariate discriminant analysis, as we have done in our earlier work. Additionally, we will use standard and newly developed statistical methods for the analysis of array data to be collected on these same samples with the aim of determining whether alternate transcripts within the classes of inflammatory and cell stress (as well as other classes) may be superior to those we have previously defined. Data from the study proposed here will be interpreted within the framework of parallel study of analysis of protein expression in these same FAD cases (by Timothy Mhyre), as well as within the framework of our almost completed (Quad) study of late onset AD (LOAD) using the same methods as proposed here. The Quad Study has four groups of ~20 each: AD, MCI, PD and control. We have also initiated a study of gene expression in peripheral leukocytes in LOAD at Sun Health Research Institute where we have access to over 150,000 elderly. PUBLIC HEALTH RELEVANCE: Alzheimer's disease is a major public health problem in the United States, as well as world-wide. In the U.S. the number of cases is 4.5 - 5 million with an annual Medicare cost of about $100 billion. As the population of the U.S. ages, the number of Alzheimer's cases is projected to grow to 16 million by 2050, with consequent immense increased expenses. The need for more effective and early diagnosis, coupled with more effective treatment that would slow or halt disease progression is clearly a pressing need. The study proposed here addresses the diagnosis arm of these needs. A number of studies have established that Alzheimer's disease has been damaging the brain for decades before symptoms become evident enough to come to medical attention. Unfortunately, by that time a large percentage of nerve cells have been lost in critical areas of the brain concerned with memory, judgment and cognitive skills. This emphasizes the importance of detecting Alzheimer's disease as early as possible so that treatment can be initiated before significant brain damage has taken place. We have obtained data that demonstrates the ability to distinguish already diagnosed Alzheimer's disease from control cases on the basis of a profile of gene expression by easily obtained peripheral blood leukocytes. The question addressed by the proposal submitted here is whether our method which has been successful at distinguishing already diagnosed disease can be successful at predicting a clinical diagnosis of Alzheimer's disease years later. In order to do this we propose to study families with gene mutations that cause early Alzheimer's disease. Some family members with such a mutation will already have AD, other family members with the mutation will not yet have been clinically diagnosed with AD, while still other family members will have neither the mutation nor Alzheimer's disease. Profiling gene expression in peripheral blood cells of these three classes of family members will tell us whether family members with the mutation but not yet showing clinically detected Alzheimer's disease show a gene expression profile similar to that of family members with the mutation who have been diagnosed with AD. If this is so, it would constitute a finding that it is possible to detect Alzheimer's disease prior to the appearance of clinically detectable symptoms in cases of familial AD, and will provide a basis for extension of these studies to the more common late onset Alzheimer's disease. [unreadable] [unreadable] [unreadable]